FR3114438A1 - Image sensor - Google Patents

Abstract

Image sensor The present description relates to an image sensor comprising a plurality of pixels (400) each comprising: a photodetector semiconductor region (110); a metal region (122) disposed on a first side of the semiconductor region; a band-pass or band-stop interference filter (160) arranged on a second face of the semiconductor region opposite the first face; and between the semiconductor region (110) and the metallic region (122), an absorbent stack (410) comprising, in order starting from the semiconductor region (110), a dielectric layer (410a), a silicon layer ( 410b) and a layer of tungsten (410c). Figure for the abstract: Fig. 4

Description

Image sensor

The present description relates generally to electronic devices, and more particularly relates to an image sensor.

Conventionally, an image sensor makes it possible to obtain images of a scene in the visible range and/or in the infrared range.

For example, the sensor captures the visible radiation coming from a scene to obtain a visible image. The image corresponds to radiation captured in one or more bands of wavelengths, for example in three bands of wavelengths corresponding respectively to the colors blue, green and red.

Some image sensors use infrared radiation to obtain a three-dimensional image of the scene. For example, the sensor is associated with an infrared pulse emitter. The travel time of the pulses from the transmitter to the scene, then from the scene to the sensor, provides depth information. The three-dimensional image is obtained from this depth information. Such a sensor is called TOF ("Time Of Flight") time-of-flight sensor. Such a sensor picks up radiation in a band of wavelengths corresponding to the wavelengths of the pulses emitted by the transmitter. This band is typically located in the near infrared, ie infrared wavelengths below 1100 nm. The sensor can supply only the depth map of the scene, or the three-dimensional image made up of the depth map combined with the visible image.

An image sensor includes a plurality of pixels generally arranged in a matrix. The radiation of each band of wavelengths of interest of the sensor is specifically picked up by pixels, distributed in the matrix, sensitive mainly to radiation in this band of wavelengths.

In practice, when a pixel is designed to be sensitive to radiation in a band of wavelengths, this pixel is also sensitive to radiation situated outside this band. Radiation located outside the target band constitutes parasitic radiation, the detection of which by the pixel reduces the quality of the image. It is therefore desirable for pixels designed to be sensitive to radiation in one of the bands of wavelengths to be as less sensitive as possible to radiation situated outside this band.

One embodiment provides an image sensor comprising a plurality of pixels each comprising:
a photodetector semiconductor region;
a metal region disposed on a first face of the semiconductor region;
a band-pass or band-stop interference filter arranged on a second face of the semiconductor region opposite the first face; And
between the semiconductor region and the metal region, an absorbent stack comprising, in order starting from the semiconductor region, a dielectric layer, a silicon layer and a tungsten layer.

According to one embodiment, the absorbing stack is capable of absorbing, in a single pass, more than 50% of incident radiation at the central wavelength of the pass band or of the cut-off band of the interference filter.

According to one embodiment, in each pixel, the semiconductor region is made of silicon.

According to one embodiment, the silicon layer has a thickness of between 20 and 100 nm.

According to one embodiment, the tungsten layer has a thickness greater than or equal to 40 nm.

According to one embodiment, the dielectric layer comprises one or more dielectric materials with refractive indices lower than that of silicon.

According to one embodiment, in each pixel, the tungsten layer is connected to a node for applying a bias potential.

According to one embodiment, in each pixel, the interference filter comprises a repetition of alternating layers of different optical indices.

According to one embodiment, in each pixel, the central wavelength is within a range of wavelengths ranging from 700 nm to 1100 nm.

According to one embodiment, each pixel comprises an additional colored or infrared filter covering the first face of the semiconductor region.

According to one embodiment, the pixels are configured to detect visible light, the sensor further comprising a plurality of depth pixels configured to detect infrared light.

According to one embodiment, the absorbing stack is present in the depth pixels.

According to one embodiment, the absorbing stack is not present in the depth pixels.

These characteristics and their advantages, as well as others, will be explained in detail in the following description of particular embodiments made on a non-limiting basis in relation to the attached figures, among which:

FIG. 1 is a cross-sectional view representing, schematically, an example of an image sensor pixel;

FIG. 2 is a diagram schematically illustrating an example of evolution, as a function of the wavelength, of the transmission of a filter of the pixel of FIG. 1;

FIG. 3 is a diagram schematically illustrating another example of evolution, as a function of the wavelength, of the transmission of a filter of the pixel of FIG. 1;

FIG. 4 is a cross-sectional view representing, schematically, an example of an embodiment of an image sensor pixel;

FIG. 5 is a diagram illustrating the parameterization of an absorbing stack of the pixel of FIG. 4

FIG. 6 is a diagram schematically illustrating an example of evolution, as a function of the wavelength, of the absorption of optical radiation by a photodetector semiconductor region;

FIG. 7 is a diagram schematically illustrating an example of evolution, as a function of wavelength, of the absorption of optical radiation by a photodetector semiconductor region in the pixel of FIG. 1 and in the pixel of FIG. 4 ;

FIG. 8 is a sectional view showing, schematically, an alternative embodiment of the pixel of FIG. 4;

FIG. 9 is a sectional view showing another alternative embodiment of the pixel of FIG. 4;

FIG. 10 is a sectional view representing, partially and schematically, an example of an embodiment of an image sensor;

FIG. 11 is a sectional view representing, partially and schematically, an example of an embodiment of a pixel of an image sensor;

FIG. 12 is a sectional view schematically representing an example of an embodiment of a pixel of an image sensor; And

FIG. 13 is a sectional view representing, partially and schematically, an example of another embodiment of an image sensor.

The same elements have been designated by the same references in the various figures. In particular, the structural and/or functional elements common to the various embodiments may have the same references and may have identical structural, dimensional and material properties.

For the sake of clarity, only the steps and elements useful for understanding the embodiments described have been represented and are detailed. In particular, elements of the image sensor such as circuits comprising transistors allowing the operation of pixels are not detailed, the embodiments described being compatible with the usual circuits allowing the operation of the pixels of an image sensor . In addition, choices of thicknesses of layers of antireflection stacks for a given wavelength, and of band-pass or band-stop filters in bands of given wavelengths, are not described in detail, the stacks and the filters described being compatible with the usual methods making it possible to choose the thicknesses of the layers of antireflection stacks and of interference filters.

Unless otherwise specified, when reference is made to two elements connected together, it means directly connected without intermediate elements other than conductors, and when reference is made to two elements connected or coupled together, it means that these two elements can be connected or be linked or coupled through one or more other elements.

In the following description, when referring to absolute position qualifiers, such as "front", "rear", "up", "down", "left", "right", etc., or relative, such as the terms "above", "below", "upper", "lower", etc., or to qualifiers of orientation, such as the terms "horizontal", "vertical", etc., it reference is made unless otherwise specified to the orientation of the figures.

Unless specified otherwise, the expressions “about”, “approximately”, “substantially”, and “of the order of” mean to within 10%, preferably within 5%.

Unless specified otherwise, the optical index denotes the real part of the complex optical index.

FIG. 1 is a sectional view representing, schematically, an example of a pixel 100 of an image sensor. The pixel is for example located in a matrix of pixels and has its side edges in contact with neighboring pixels.

In the orientation of the figure, the sensor is intended to receive optical radiation arriving from the top of the pixels. Optical radiation is referred to here as visible radiation or radiation located in the near infrared. The optical radiation typically comes from an element of a scene located opposite the pixel.

Pixel 100 includes a semiconductor photodetector region 110, typically made of silicon. The semiconductor region 110 can be delimited by isolation trenches 115 filled with an electrical insulator, for example silicon oxide. As a variant, the trenches comprise conductors insulated from the region 110. The trenches 115 separate the various semiconductor regions 110 from the neighboring pixels of the image sensor.

By way of example, the pixel extends laterally over a width comprised between 1.2 and 5 micrometers, for example of the order of 1.4 micrometers. Region 110 typically has a thickness, corresponding to the height in the orientation of the figure, of between 2 and 10 micrometers, for example of the order of 6 micrometers.

The semiconductor region 110 has its underside, or front face, covered with an insulating layer 120. Metallic regions 122 are located in the layer 120. The regions 122 are typically conductive elements such as conductive traces and/or conductors. filling in vias. These conductive elements interconnect components not shown, such as transistors, which allow the operation of the pixel and/or of various parts of the image sensor.

In the example represented, the pixel comprises a lens 130 on the side of arrival of the optical radiation. The lens 130 makes it possible to focus the optical radiation towards the photodetector region 110.

Furthermore, the pixel 100 can comprise a filter 140 covering the face of the photodetecting region 110 located on the side of arrival of the optical radiation. Filter 140 is preferably located between lens 130 and photodetector region 110. Filter 140 serves to select the wavelengths of radiation detected by the pixel. Preferably, the filter 140 is an organic filter, for example made of resin. The filter 140 can constitute an infrared filter, that is to say allowing optical radiation to pass in infrared wavelength bands, preferably in the near infrared. The filter 140 can constitute a colored filter, that is to say allowing optical radiation to pass in a band of wavelengths in the visible range, preferably corresponding to the color red, green or blue.

The pixel may further include an antireflection layer 150. Layer 150 may include one or more dielectric layers. The antireflection layer 150 is provided to limit the upward reflection of optical radiation at the wavelengths detected by the pixel.

Pixel 100 further comprises an interference filter 160. An interference filter consists of a stack of several layers of alternating optical indices. By way of example, the interference filter comprises, preferably consists of, an alternation of first and second layers, the first layers being identical to each other, and the second layers being identical to each other. Preferably, the interference filter comprises at least two, for example at least three, first layers and at least two, for example at least three, second layers. By way of example, the first layers are dielectric, and the second layers are dielectric and/or are sufficiently thin metal layers to be at least partially transparent. The interference filter preferably consists of layers of silicon oxide, between which are interposed layers of silicon, or titanium oxide, or silicon nitride. The thickness of the interference filter is preferably less than the width of the region 110, for example less than 2.0 micrometers. The interference filter 160 can be band-stop or band-pass, that is to say respectively block or allow radiation to pass in a band of wavelengths.

FIG. 2 is a diagram illustrating schematically, by way of example, the evolution as a function of the wavelength λ (on the abscissa, in nm) of the transmission T200 (on the ordinate, between 0% and 100%) of the interference filter 160 of the pixel of FIG. 1, in the case where the filter 160 is a notch filter. The transmission is defined, for radiation arriving on the upper face of the filter 160 and substantially orthogonal to the mean plane of the layers of the filter 160, as being equal to the percentage of this radiation which exits via the lower face of the filter.

The band-stop interference filter is designed to block radiation in the near infrared, around a central wavelength λ0 for example between 700 and 1100 nanometers (of the order of 875 nm in the example shown). The central wavelength λ0 corresponds for example to a transmission minimum. Preferably, the filter 160 passes less than 30% of the radiation at the wavelength λ0. Preferably, the interference filter is designed so that the transmission is weak, for example less than 50%, in a range of wavelengths 210, and strong, for example greater than 75%, outside a range. of wavelengths 220 including range 210. Range 210 defines the cutoff band of the filter. By way of example, pad 210 has a width greater than 150 nm and is entirely between 700 and 1100 nanometers, and pad 220 has a width greater than 250 nm. The difference between the widths of the pads 220 and 210 is for example between 100 and 200 nm.

In this example, the interference filter 160 comprises seven alternating layers of silicon oxide and silicon nitride. The total thickness of the filter is on the order of 1.0 micrometer.

Preferably, such a band-stop interference filter is used in the near infrared when the pixel is designed to detect visible light. The optional filter 140 is then a colored filter. Although the filter 140 is designed to pass only radiation from a band of wavelengths in the visible range, in practice, when the filter 140 is for example made of resin, the filter 140 also passes radiation in the near infrared. The interference filter 160 then aims to limit the detection of radiation in the near infrared.

The inventors have observed that in fact, despite the presence of the interference filter 160, the radiation in the band of wavelengths partially cut off by the interference filter 160 remains detected by the detector region 110. As a result, in the sensor of images, the detection of near infrared radiation is added to the detection of visible radiation corresponding to the image, which reduces the quality of the image.

FIG. 3 is a diagram illustrating schematically, by way of example, the evolution as a function of the wavelength λ (on the abscissa, in nm) of the transmission T300 (on the ordinate, between 0% and 100%) of the interference filter 160 of the pixel of FIG. 1, in the case where the filter 160 is a band-pass filter.

The band-pass interference filter is designed to pass radiation in a band of wavelengths around a central wavelength λ0. The central wavelength is preferably located in the near infrared, for example between 700 and 1100 nm (of the order of 950 nm in the example shown). The transmission preferably has the shape of a peak around the central wavelength. The transmission is for example close to 100% for the central wavelength λ0, and decreases on either side of the wavelength λ0. The width L at mid-height of the peak is typically around 35 nm. The peak can have multiple vertices (two vertices in the example shown) and the mid-height is defined by half the height of the highest vertex. The transmission band, or bandwidth, of the filter is defined by the wavelengths for which the transmission is greater than half the height of the peak. The central wavelength λ0 then corresponds to the center of the passband. The transmission is close to zero, for example less than 1%, at more than 50 nm from the central wavelength λ0.

The filter 160 comprises in this example eighteen alternating layers of silicon and silicon oxide. The total thickness of the filter is for example of the order of 1.5 micrometers.

Such a band-pass filter is used for example when the sensor is of the time-of-flight type, associated with an infrared pulse emitter. The filter is then designed so that the wavelengths of the pulses are located in the passband of the filter. For example, the wavelength of the pulses is located in a band 310 centered on the wavelength λ0. By way of example, the strip 310 has a width of the order of 25 nm. Preferably, the filter 140 then allows only infrared radiation to pass, which makes it possible to cut off any secondary peaks that the filter 160 could present outside the range of infrared radiation. Alternatively, filter 140 can be omitted.

The inventors have observed that in practice, the photodetector region 110 detects radiation in a band of wavelengths wider than the desired transmission peak. As a result, the pixel detects, in addition to the pulses, infrared radiation situated outside the band 310. These infrared radiation constitute parasitic radiation which is detrimental to the quality of the depth map obtained.

The embodiments described below make it possible to improve the quality of the images obtained from visible radiation, and/or the quality of the depth images.

FIG. 4 is a cross-sectional view representing, schematically, an example of an embodiment of an image sensor pixel 400 .

Pixel 400 comprises elements identical or similar to those of pixel 100 of FIG. 1, arranged in an identical or similar manner, namely:
- a photodetector semiconductor region 110, for example delimited by isolation trenches 115;
- one or more metal regions 122;
- an interference filter 160; And
- preferably, a lens 130 and/or a filter 140, and/or an antireflection layer 150.
These elements are not described here again in detail.

The pixel 400 comprises, between the metallic region or regions 122 and the photodetector region 110, an absorbent structure 410. Preferably, the pixel does not comprise metallic elements between the photodetector region 110 and the absorbent structure 410.

The absorbent structure 410 is constituted by a stack of layers comprising, in order starting from the lower face of the photodetecting region 110, a dielectric layer 410a, a silicon layer 410b, and a tungsten layer 410c. Layer 410a is for example a layer of silicon oxide or a layer of silicon nitride. As a variant, layer 410a can be formed by a stack of several layers of dielectric materials with refractive indices lower than that of silicon, for example one or more layers of silicon oxide and one or more layers of nitride of silicon. The dielectric layer 410a is for example in contact, via its upper face, with the lower face of the photodetecting region 110. The layer 410a makes it possible in particular to electrically insulate the layers 410b, 410c and the conductive elements 122 from the photodetecting region 110. The Silicon layer 410b is for example in contact, via its upper face, with the lower face of dielectric layer 410a. The tungsten layer 410c can be in contact, via its upper face, with the lower face of the silicon layer 410b. As a variant, a thin tie layer, for example made of titanium nitride, for example with a thickness of less than 10 nm, can interface between the silicon layer 410b and the tungsten layer 410c.

The thicknesses of the layers 410a, 410b and 410c of the stack 410 are chosen so that the stack 410 has, for the central wavelength λ0 of the interference filter 160, an absorption coefficient greater than that of the material semiconductor of the region 110. The stack 410 is dimensioned so that, for the central wavelength λ0 of the interference filter 160 and for an incident radiation substantially orthogonal to the mean plane of the stack 410, more than 50%, of preferably more than 80%, preferably more than 95%, of the radiation entering stack 410 is absorbed in stack 410 in a single pass. In other words, more than 50%, preferably more than 80%, preferably more than 95%, of radiation entering through the upper face of the stack 410 is absorbed in the stack 410 and is not reflected towards the region 110. For example, for the wavelength λ0, approximately 90% of the radiation entering the stack 410 is absorbed in the stack 410 in a single pass. For example, more than 50%, preferably more than 80%, preferably more than 90% of all radiation in the wavelength range between 900 nm and 1000 nm is absorbed in a single pass through stack 410. Preferably, more than 50%, preferably more than 80%, preferably more than 90% of all radiation in the cutoff band or in the passband of the filter 160 is absorbed in a single pass through the stack 410.

If we denote respectively by N1 the refractive index of the layer 410a, by N2 the refractive index of the layer 410b and by N3 the refractive index of the layer 410c, and by considering the thicknesses of the layers 410a and 410c as infinite (or more precisely semi-infinite), it can be shown that the reflection of the stack 410 is zero if the following equation holds: with where d is the thickness of layer 410b.

If, moreover, N3 comprises a non-zero imaginary part, then the transmission is zero and the wave is completely absorbed by the stack 410.

The studies carried out by the inventor have shown that by taking N1 equal to the refractive index of silicon oxide and N2 equal to the refractive index of silicon, and for a wavelength λ=940 nm , the choice of tungsten as the material of the 410c layer (index N3) makes it possible to approach as closely as possible the equality of the equation [Math 1] above. A thickness of silicon d=39 nm then makes it possible to obtain maximum absorption (close to 100%) of the radiation by the stack 410.

FIG. 5 illustrates the relevance of the choice of tungsten compared to other metals usually used in integrated circuits.

Considering the aforementioned equation [Math 1], FIG. 5 is a diagram illustrating the evolution of the complex optical index N3 when the thickness d of the silicon layer 410b varies from 0 to 130 nm. More particularly, in the diagram of FIG. 5, the abscissa axis represents the real part Real(N3) of the index N3, and the ordinate axis represents the imaginary part Imag(N3) of the index N3. The circular curve 501 of FIG. 5 represents the evolution of the index N3 (real part and imaginary part) when the thickness d varies from 0 to 130 nm. Point N0 of curve 501 corresponds to the value taken by index N3 for d=0 and for d=130 nm. For these thicknesses, the imaginary part of the index N3 is zero.

A point N _W corresponding to the complex optical index of tungsten, a point N _Cu corresponding to the complex optical index of copper, a point N _Ag corresponding to the optical index complex of silver, and a point N _Al corresponding to the complex optical index of aluminum.

As this appears in the figure, the point N _W almost coincides with a point of the circle 501 having a non-zero imaginary part. This point corresponds to the value of the index N3 for a thickness d of silicon equal to 39 nm. The points N _Cu , N _Ag and N _Al are for their part very far from the circle 501. This shows that, among the metals mentioned above, tungsten is the only relevant candidate for obtaining the desired absorption effect.

In practice, the thicknesses of the layers 410a and 410c are of course not semi-infinite. Additionally, as indicated above, dielectric layer 410a may include one or more dielectric materials other than silicon oxide. Numerical simulations make it possible to adjust the thickness d of silicon to be provided in order to maximize absorption, as a function of the various parameters of the stack 410 and/or of the central wavelength λ0 of the interference filter 160. As an For example, the thickness d of the silicon layer 410b is between 20 and 100 nm, and preferably between 30 and 50 nm.

Furthermore, to maximize the absorption in the stack 410, the thickness of the tungsten layer 410c is preferably relatively high, for example greater than 40 nm and preferably greater than 60 nm.

In the case where the interference filter is a notch filter of the type described in relation to FIG. 2, the presence of the absorbing structure 410 makes it possible to reduce the detection of parasitic radiation in the near infrared, and thus to improve the quality of the image.

In the case where the interference filter is a band-pass filter of the type described in relation to FIG. 3, among the infrared rays detected, the proportion of infrared rays situated outside the band 310 is lower in the pixel of 4 than in the pixel of FIG. 1. The presence of the structure 410 therefore reduces the proportion of parasitic radiation detected and thus makes it possible to improve the quality of the depth image captured.

An explanation of the function of layer 410 is detailed below in relation to Figure 6.

FIG. 6 is a diagram illustrating schematically, by way of example, the evolution as a function of the wavelength λ (on the abscissa, in nm) of the absorption T600 of the optical radiation by the photodetector region 110, in one single pass through the photodetector region 110.

Radiation located in the near infrared is partially absorbed and detected by region 110. For example, for the central frequency λ0 of the interference filter, region 110 absorbs and detects in a single pass typically between 1 and 50%, preferably between 2 and 25% of the radiation that arrives in region 110.

In the absence of the absorbing structure 410, radiation arriving in the region 110 can cross the region 110 a first time from top to bottom almost without being absorbed, then a second time from bottom to top after having been reflected on the metal regions 122. The radiation thus performs a round trip.

For certain wavelengths, the interference filter 160 allows only a part, for example between 1% and 80%, of the radiation to pass. This occurs for example for the whole band of wavelengths cut by the notch filter of figure 2, and in the lateral parts of the transmission peak of the bandpass filter of figure 3. The interference filter then reflects the non-transmitted part of the radiation.

In the absence of the absorbent structure 410, for these wavelengths of partial transmission by the filter 160, the radiation having passed through the filter 160 then performs multiple round trips between, at the bottom, the metal regions 122, and, at the top, the partially reflective filter 160. The less the interference filter allows the radiation to pass, the more it is reflective and the more the number of round trips increases. Thus, the reduction by the filter 160 of the radiation reaching the region 110, is accompanied by an increase in the number of passages in the region 110. Once entered in the region 110, radiation is all the more detected that the number of round trips is high. Accordingly, the reduction of incoming radiation by filter 160 does not result in a reduction of the same magnitude in the detection sensitivity of the pixel.

In the presence of the absorber stack 410, the number of times radiation passes through the region 110 is limited by the stack 410. This reduces the radiation detection sensitivity for the partial transmission wavelengths of the filter 160. of the absorbing stack 410, the partial blocking by the filter 160 of the parasitic radiation then results in a corresponding reduction in the sensitivity of detection by the pixel of the parasitic radiation. The quality of the visible and/or infrared image has thus been improved.

FIG. 7 is a diagram schematically illustrating an example of evolution, as a function of the wavelength λ (in nm, on the abscissa), of the absorption A (on the ordinate, from 0 to 1) of optical radiation by the photodetector semiconductor region 110 in the pixel of FIG. 1 (curve 601 of FIG. 7) and in the pixel of FIG. 4 (curve 603 of FIG. 7).

Strong oscillations can be observed on curve 601 for wavelengths beyond approximately 700 nm. These oscillations illustrate the effect described above in relation to FIG. 6. More particularly, in the pixel of FIG. 1, the semiconductor region 110 forms a cavity for the radiations close to the central wavelength of the interference filter 160 This cavity exhibits resonances at certain frequencies, which explains the oscillations of curve 601. This cavity effect degrades the quality of the acquired images.

On curve 603, it is observed that the oscillations are greatly reduced. This results from the provision of the absorbing structure 410 which makes it possible to avoid the resonant cavity effect described above.

Although a silicon photodetector region has been described here, partially detecting radiation in the near infrared, the photodetector region could be made of another semiconductor. This other semiconductor can then have a range of wavelengths in which it absorbs and detects radiation only partially. The central wavelength of the interference filter is chosen within this range. Preferably, the absorbent structure 410 is sized to absorb in a single pass greater than 50%, for example greater than 80%, of all radiation in that wavelength range.

FIG. 8 is a cross-sectional view representing, schematically, an alternative embodiment 800 of the pixel of FIG. 4.

In this variant, the tungsten layer 410c of the absorbent stack 410 is connected, preferably connected, to a node for applying a bias potential VB, for example via one of the metal regions 122 constituting an interconnecting track. The application of a polarization potential to the absorbent stack 410 makes it possible to evacuate any electric charges trapped in the insulating layer 410a. Furthermore, the application of this bias potential can make it possible to avoid storing charges in the layers 410b and/or 410c, which could disturb the electric potential of the region 110.

By way of example, the isolation trenches 115 each comprise a conductive region 810 insulated from the region 110 by an electrically insulating material 815. The regions 810 are connected, preferably connected, to some of the metal regions 122 by connections 820 There was thus obtained, for each trench, a capacitive element making it possible to electrostatically influence the region 110, and/or to reduce or neutralize the dark current. Indeed, the reduction or the neutralization of the dark current can be obtained by accumulation of holes to store electrons, or of electrons to store holes.

The stack 410 has openings 830 at the level of the connections 820. The openings 830 make it possible to electrically isolate the connections 820 from the absorbent stack 410. Any electrical connection passing through the stack 410 can be isolated from the layer 410 by an opening 830. These electrical connections form, for example, contacts to pixel elements such as transistors or junctions. The openings 830 are preferably located at the edges of the pixel. However, regions 810, leads 820, and openings 830 can be omitted.

FIG. 9 is a cross-section view schematically representing another embodiment variant 1300 of the pixel of FIG. 4. In this variant, cavities 1310 extend into region 110 from its upper face. The cavities 1310 are filled with a dielectric material, for example silicon oxide. The cavities 1310 are preferably arranged in a network. The pitch of the grating is for example greater than half of a central wavelength of the band of wavelengths detected by the pixel. In the case where the filter 160 is a band-pass filter, the pitch of the grating is preferably greater than half the central wavelength λ0 of the filter 160.

The cavities 1310 and the portions 1320 of the region 110 arranged between these cavities constitute a diffraction grating. This diffraction grating introduces angles into the propagation of the radiation entering from the top of the photodetector region 110. The radiation then performs several round trips with a horizontal component in the region 110 between the side walls of the region 110. This increases the length of the path taken by the radiation inside the region 110, and therefore the probability that this radiation will be detected by the region 110. Unlike the round trips between the bottom and the top of the pixel, the number of which depends on the transmission of the filter 160, the horizontal round trips increase the detection substantially in the same way for all the wavelengths of the band detected by the pixel. Thus, the diffraction grating in the upper part of the region 110 gives the pixel 1300 a higher sensitivity to radiation than that of a pixel without this diffraction grating, while retaining the advantage of the image quality conferred by the limitation the number of vertical round trips.

FIG. 10 is a sectional view representing, in a partial and schematic manner, an example of an embodiment of an image sensor 1400 comprising pixels of the type of pixel 400 of FIG. 4. The sensor 1400 can capture three-dimensional color images or color and infrared images.

More specifically, a group of four pixels 400R, 400G, 400B, and 400Z, of sensor 1400 has been represented. Pixels 400R, 400G, 400B, and 400Z each correspond to pixel 400 of FIG. constituted by a filter respectively 140R, 140G, 140B, and 140Z, respectively allowing blue light, green light, red light, and radiation in the near infrared to pass. The 140Z filter is optional.

In each of the pixels 400R, 400G and 400B, the interference filter 160 of the pixel 400 of FIG. 4 consists of a band-stop interference filter 160BC, of the type of the filter of FIG. 2. Preferably, the filter 160BC is the same continuous filter common to pixels 400R, 400G and 400B. In the pixel 400Z, the interference filter 160 of the pixel 400 of figure 4 consists of a band-pass interference filter 160BP, for example of the type of the filter of figure 3.

In the example, the absorbing stack 410, and the layer 120 are common for the various pixels of the sensor, and in particular common to the four pixels of the group of pixels represented.

As mentioned above, the absorbing stack 410 limits the number of round trips made by the light between the top and the bottom of each of the pixels, thus improving the quality of the captured image.

Furthermore, the absorbing stack 410 limits the quantity of optical radiation, in particular infrared, which, after having crossed from top to bottom the region 110 of one of the pixels, is reflected by the metallic regions 122 towards the neighboring pixels. This corresponds to a further improvement in image quality.

To produce the pixels 400R, 400G, 400B, and 400Z, the regions 110, and the trenches 115 separating the regions 110, are preferably formed beforehand in a semiconductor substrate. The layers 410a, 410b and 410c of the absorbent stack 410, and the insulating layers 120 containing the metal regions 122 are then successively formed on the front face of the substrate (lower face). Preferably, a handle 1410 is then glued , for example a semiconductor wafer, on the front face of the insulating layers 120. All the elements are removed from the rear face of the substrate up to a level defining the upper level of the regions 110, for example by polishing.

Then, in a first step, the filters 160BP are formed. To do this, for example, all of the upper faces of the regions 110 are covered with a first stack of alternating layers forming the filter 160BP, then the portions of the first stack of layers located at the locations of the 400R, 400G and 400B pixels.

In a second step, the filters 160BC are formed. For this, for example, the structure obtained in the first step is covered with a second stack of layers corresponding to the filters 160BC. The portions of the second stack located at the location of the pixel 400Z are then removed, for example by mechanical-chemical polishing.

The filters 140R, 140G and 140B are then formed, and the possible filter 140Z, then the possible lenses 130.

The parasitic radiations that the filter allows to pass partially are all the more numerous as the filter is thin. Due to the fact that the absorbing stack 410 reduces the detection of these parasitic radiations, it is possible, for a given quantity of parasitic radiations absorbed, that is to say for a given image quality, to reduce the thickness of the filter by compared to a sensor not comprising a layer 410A. It is then possible to bring the lenses 130 closer to the regions 110, which makes it possible to increase the viewing angle of the sensor, without altering the image quality. Furthermore, this reduction in thickness corresponds to a reduction in the width, or horizontal dimension, of a region 1420 in which the second stack, which corresponds to the filters 160BC, has its layers stacked laterally against the side of the filter 160BP. In region 1420, filter 160BC has degraded filtering properties. Reducing this region makes it possible to reduce the size of the pixels without altering the image quality, compared to a sensor without an absorbent layer 410.

In the example of FIG. 10, the absorbing stack 410 extends under the pixels 400R, 400G and 400B and under the pixel 400Z. As a variant, the stack 410 can extend only under the pixels 400R, 400G and 400B and not extend under the pixels 400Z. In this case, the manufacturing method is for example similar to what has just been described, except that a step of localized withdrawal of the absorbent stack 410 vis-à-vis the pixels 400Z is provided before the step of depositing the insulating layers 120.

FIG. 11 is a cross-sectional view representing, in a partial and schematic manner, an example of an embodiment of an image sensor 1450 comprising pixels of the type of pixel 400 of FIG. 4. The sensor 1450 repeats the elements of the sensor 1400 of FIG. 10, with the difference that the manufacture of the filters 160BC and 160BP includes an additional step between the first and second steps.

At this additional step, a dielectric layer, for example of silicon oxide, is formed at the locations of the pixels 400R, 400G and 400B, up to the upper level of the filter 160BP formed in the first step. To do this, for example, the structure obtained in the first step is covered with the oxide layer, then all the elements situated above the upper face of the filter 160BP are removed, for example by mechanical-chemical polishing.

Then, in the second step, the second stack, which corresponds to the filter 160BC, is deposited on the flat surface left by the polishing, and the portions of the second stack are removed from the location of the pixel 400Z by etching.

Advantageously, since the filters 160BC are formed on a flat surface and are delimited by etching, they form a stack of flat layers of constant thickness over the entire surface of the filter. The quality of the filtering is thus better than with a filter consisting of non-flat layers and/or of non-constant thicknesses.

At a later stage, it is possible to form, at the location of the pixel 400Z, a portion of dielectric layer, for example of silicon oxide, up to the upper level of the filter 160BC. To do this, for example, the structure obtained in the second step is covered with a dielectric layer, for example of silicon oxide, then all the elements located above the upper face are removed, for example by mechanical-chemical polishing. of the 160BC filter.

Here again, although the absorbing stack 410 has been shown extending under the pixels 400R, 400G and 400B and under the pixel 400Z, as a variant, the stack 410 can only extend under the pixels 400R , 400G and 400B and not extend below 400Z pixels.

FIG. 12 is a schematic sectional view, representing an example of an embodiment of a pixel 1500 of an image sensor. Pixel 1500 comprises elements identical or similar to those of pixel 400 of FIG. 4. Pixel 1500 differs from that of FIG. 4 in that it comprises, between absorbing stack 410 and insulating layer 120, successively from the lower face of the absorbent stack 410, a dielectric layer 1510, a conductive region 1520, an insulating layer 1525, and a semiconductor region 1530, for example made of silicon. Conductive region 1520 and insulating layer 1525 can be omitted.

To produce pixel 1500, region 110 and trenches 115 are formed in a semiconductor substrate. The front face of the substrate (lower face) is then covered by the absorbent stack 410. The tungsten layer 410c of the absorbent stack 410 is covered by the dielectric layer 1510, for example made of silicon oxide.

The dielectric layer 1510, the conductive region 1520 is then formed. For example, the region 1520 defines a ground plane and covers the whole of the lower face (or front) of the semiconductor substrate. The insulating region 1525 is then formed, for example in silicon oxide. After that, semiconductor region 1530 is bonded to layer 1525 by molecular bonding. Electronic circuit components of the image sensor are then formed, for example transistors (not shown) in and on the semiconductor region 1530, on the side of its lower face (front face). Then the insulating layers and the metallic regions 122 are formed. The regions 122 preferably correspond to interconnection tracks between the components of the circuits of the image sensors.

The layer 150, the interference filter 160, the possible filter 140 and the possible lens 130, are for example formed at a later stage of the manufacture of the pixel.

An advantage of pixel 1500 is that part of the circuits of the image sensor are formed in and on region 1530 independently of photodetector regions 110, which makes it possible to reduce the size of the pixels or to integrate additional functions therein, for compared to an image sensor not comprising the semiconductor region 1530. In addition, the presence of the region 1520 can make it possible to optimize the operation of the circuits of the image sensor.

The presence of layer 1510 and region 1520 in pixel 1500 has the same effect on light reflection as layer 120 and regions 122 in pixel 400 of Figure 4. In pixel 1500, the stack 410 limits the number of round trips that the radiation can make in the cavity 110 after reflection on the region 1520 and on the filter 160. In a sensor comprising pixels of the type of the pixel 1500, an image quality is obtained improved, in the same way as in an image sensor comprising pixels of the type of pixel 400 of FIG.

FIG. 13 is a sectional view representing, in a partial and schematic manner, an example of an embodiment of an image sensor 1600. The image sensor 1600 comprises elements identical or similar to those of the sensor 1400 of FIG. 11, arranged in an identical or similar manner, with the difference that the 400Z pixel is replaced by a 1600Z pixel.

The pixel 1600Z comprises elements identical or similar to those of the pixel 400Z of FIG. 11, arranged in an identical or similar manner, namely a photodetecting region 110Z, an absorbing stack 410Z comprising, in order from the the photodetector region 110Z, a dielectric layer 410Za, a silicon layer 410Zb and a tungsten layer 410Zc, and insulating layers 120Z. The photodetector region 110Z, the stack 410Z and the insulating layers 120Z are identical or similar to respectively the photodetector region 110, the absorbing stack 410 and the insulating layers 120 of the pixel 400Z of FIG. 4. In the pixel 1600Z, the region Semiconductor photosensor 110Z is located at a lower level than photosensor regions 110 of pixels 400R, 400G, and 400B. Pixel 1600Z further includes optical transmission regions 1602 and 1604 located between its photodetector region 110Z and its filter 160BP. It will be noted that in the example represented, in vertical projection, the outline of the photo-detector region 110Z substantially coincides with the outline of the optical transmission regions 1602 and 1604. As a variant, the photo-detector region 110Z can be extend, laterally, beyond the optical transmission regions 1602 and 1604.

Region 1602 is located at the same level as regions 110 of pixels 140R, 140G and 140B, and region 1604 extends vertically between region 110Z and optical transmission region 1602. Region 110Z of pixel 1600Z is located in a substrate 1610 which extends horizontally under layer 120 common to pixels 400R, 400G and 400B. The absorbent stack 410Z and the insulating layers 120Z extend horizontally on the underside of the substrate 1610.

Components such as transistors, not shown, are formed on the lower face, or front face, of substrate 1610. These transistors are for example interconnected by tracks constituting metal regions 122Z located in layer 120Z. The transmission region 1602 is preferably made of the same semiconductor as the regions 110 of the pixels 400R, 400G, 400B. Region 1604 is preferably made of a dielectric, for example silicon nitride, or amorphous silicon, and extends from the underside of region 1602, through absorbent stack 410 and insulating layers 120 .

In operation, the optical radiation transmitted by the filter 160BP is guided by the optical regions 1602 and 1604 to the photodetector region 110Z.

The sensor may have densities of transistors, and metal interconnects formed by regions 122 and 122Z between these transistors, greater than the densities of transistors and interconnects in a sensor lacking a 1610 substrate extending below pixels 400R , 400G and 400B. The 1600 sensor can thus be particularly compact.

To produce the sensor 1600, the regions 110, and the trenches 115 separating the regions 110, are formed in a semiconductor substrate 1620, for example made of silicon. Then, on the front face of the substrate 1620 (lower face), any components such as transistors, the absorbing stack 410, the insulating layers 120 containing the metallic regions 122, and the region 1604 are formed.

On the front face of the substrate 1610 (lower face) comprising the regions 110Z, optional components such as transistors, the absorbent stack 410Z, and the insulating layers 120Z containing the metal regions 122Z are formed separately.

A handle 1630, for example a semiconductor substrate, is glued onto the front face of the insulating layer 120Z. The parts of the substrate 1610 located on the side of the rear face are then removed up to the upper level of the regions 110Z, for example by mechanical-chemical polishing.

The rear face of the substrate 1610 is then glued to the front face of the layer 120 and of the region 1604.

All the elements are then removed from the rear face of the substrate 1620 up to a level defining the upper level of the regions 110 of the pixels 400R, 400B and 400Z, for example by polishing.

The filters 160BC, 160BP, 140R, 140G, 140B, the possible filter 140Z and the possible lenses 130 are then formed, in the same way as that described in relation to FIG. 11 to form these elements.

Alternatively, absorbent stack 410Z may be omitted from the structure of Figure 13.

Various embodiments and variants have been described. Those skilled in the art will understand that certain features of these various embodiments and variations could be combined, and other variations will occur to those skilled in the art. In particular, the variants and embodiments described in relation to FIGS. 8 and 9 are mutually compatible and can be applied, alone or in combination, to the embodiments of FIGS. 10 to 13.

In the embodiments described above, the lower layer 410c of the absorbent stack 410 preferably corresponds to a specific level of tungsten, that is to say deposited only for the production of the absorbent stack 410. However , as a variant, the lower layer 410c of the absorbent stack 410 may correspond to a first level of metallization of the sensor, in which interconnection metallizations are also formed at the periphery of the pixels.

Finally, the practical implementation of the embodiments and variants described is within the reach of those skilled in the art based on the functional indications given above.

Claims (13)

Image sensor comprising a plurality of pixels (400) each comprising:
a photodetector semiconductor region (110);
a metal region (122) disposed on a first side of the semiconductor region;
a band-pass or band-stop interference filter (160) disposed on a second face of the semiconductor region opposite the first face; And
between the semiconductor region (110) and the metal region (122), an absorbent stack (410) comprising, in order starting from the semiconductor region (110), a dielectric layer (410a), a silicon layer (410b ) and a layer of tungsten (410c). Image sensor according to Claim 1, in which the absorbing stack (410) is capable of absorbing, in a single pass, more than 50% of incident radiation at the central wavelength (λ0) of the strip bandwidth or cut-off band of the interference filter. An image sensor according to claim 1 or 2, wherein in each pixel the semiconductor region (110) is silicon. Image sensor according to any one of Claims 1 to 3, in which the layer of silicon (410b) has a thickness of between 20 and 100 nm. Image sensor according to any one of Claims 1 to 4, in which the tungsten layer (410c) has a thickness greater than or equal to 40 nm. An image sensor according to any of claims 1 to 5, wherein the dielectric layer (410a) comprises one or more dielectric materials with refractive indices lower than that of silicon. Image sensor according to any one of Claims 1 to 6, in which, in each pixel, the tungsten layer (410c) is connected to a node (VB) for applying a bias potential. An image sensor according to any one of claims 1 to 7, wherein in each pixel the interference filter (160) comprises a repetition of alternate layers of different optical indices. An image sensor according to any one of claims 1 to 8, wherein in each pixel the center wavelength (λ0) is within a wavelength range from 700 nm to 1100 nm. Image sensor according to any one of Claims 1 to 9, in which each pixel comprises an additional colored (140R, 140G, 140B) or infrared (140Z) filter covering the first face of the semiconductor region (110). A sensor according to any of claims 1 to 10, wherein said pixels (400R, 400G, 400B) are configured to detect visible light, the sensor further comprising a plurality of depth pixels (400Z) configured to detect infrared light. A sensor according to claim 11, wherein the absorbent stack (410) is present in the depth pixels (400Z). A sensor according to claim 11, wherein the absorbent stack (410) is not present in the depth pixels (400Z).